Codebase audit on GPT6 Astra high used just 41% of quota in 30+ minutes
Yamapama · x · 2026-09-05
A user shares token-usage observations from running codebase audits on GPT6 Astra high: a full audit took over half an hour while staying at 41% usage, and a prior audit stayed under 1% usage throughout, surfacing several bugs and optimizations now being implemented.
More from Models
- GPT-6 in Codex still needs human steering on hard problems, researcher reports — DimitrisPapail · 2026-09-05
- Conditional independence limits parallel token generation in masked diffusion models — alec_helbling · 2026-09-05
- GPT-6 Astra guide ships an official prompt to curb the model's excessive test-running habit — JeremyNguyenPhD · 2026-09-05
- Fine-tuned Qwen3-8B on a single RTX 4090 produces text that scores 100% human on pangram — StewartalsopIII · 2026-09-05
- 27B open-weight model turns a dimension sketch into editable FreeCAD parts — MaziyarPanahi · 2026-09-05
- Astra system card reveals Apollo testing lasted just three days with CoT access — tallinzen · 2026-09-05